The Application and Use of an Automated Spatial Pattern Recognition (SPR) System in the Identification and Solving of Yield Issues in Semiconductor Manufacturing

Modern semiconductor manufacturing facilities process thousands of wafers a week and produce enormous volumes of data that must be quickly and efficiently mined for yield signals. AMD has embarked on a project with Rudolph Technologies to create an automated pattern recognition system for use in AMD manufacturing and test facilities. This paper will explain what advantages are gained by deploying an automated spatial pattern classification system, as well as some of the challenges in making the system successful. We will also give examples of how SPR information is used for control, problem identification, and ultimately root cause determination in a high volume fab environment.